Which AI chatbots have ads? 3 sell, 3 do not (2026)

Which AI chatbots have ads right now?

As of 26 October 2026, three AI surfaces sell ads: ChatGPT, Google's AI answers in Search, and Microsoft Copilot. The Gemini app, Perplexity and Claude sell nothing. Grok, Meta AI and Amazon's Rufus are outside this map.

Only one of the three needs new money. If you run broad match, AI Max or Performance Max on Google in one of its 12 eligible English-language markets, your ads are probably inside AI Overviews already, and Google Ads cannot tell you which ones. Below: what is buyable in our five markets, and a calculator that tells you whether the new-money test paid.

The newest changes: on 21 October 2026 OpenAI added 26 countries, taking ChatGPT Ads to 89, and on 5 October it announced a visual ad format for image generation. Every row was checked on 26 October; Perplexity rests on press reporting.

Read the "how you buy it" column first: "Live" covers two buying models.

Two buying routes, and only one needs new budget

Route 1 is a new ad platform: ChatGPT advertising, with its own account and policy review. Route 2 is search inventory rendered inside an AI answer, reached through the Google Ads and Microsoft Advertising campaigns you already run.

Route 1 is a new line item: you open an OpenAI Ads Manager account and set a budget, and OpenAI decides which ad, if any, sits under an answer. Nothing you run on Google or Microsoft reaches it.

Route 2 has no separate buy, but it is work. Google's help page says "the user query as well as the content of the AI Overview are considered when serving these ads", so your feeds and landing pages decide whether you qualify; advertisers cannot opt out of AI Overview placements or target them directly. Reach depends on your matching: broad match, AI Max for Search or Performance Max.

Can you run ChatGPT ads in Singapore, Malaysia, Australia and North America?

Yes. OpenAI's Ads Manager availability page, read on 26 October 2026, lists 89 countries as available for self-serve, Singapore, Malaysia, Australia, Canada and the United States among them.

The billing rule matters more than the market list: "the legal entity that will advertise and be billed must be based in a country listed as available." India is listed. Hong Kong is not.

Most had no self-serve access when we covered OpenAI's advertising plans in July. OpenAI says the test began in the US on 9 February 2026; it announced pilots for Canada, Australia and New Zealand on 26 March, launched in the UK, Mexico, Brazil, Japan and South Korea on 11 August, and opened self-serve across India, Europe, the Middle East and North Africa on 31 August.

On 23 September it announced Singapore, Malaysia, Indonesia, the Philippines, Thailand, Vietnam and Taiwan, taking it to "over 60 countries". On 5 October it announced a visual ad format shown while ChatGPT generates images, with US testing to start later in October. On 21 October it added 26 more, "extending availability to 89 countries worldwide".

Across the five markets the picture on 26 October 2026 was the same: ChatGPT self-serve open, ads inside AI Overviews except in sensitive verticals, finance and health generally prohibited on ChatGPT outside the US, where category rules apply instead. Only the US has a Google-announced AI Mode ads test.

• One ad wins each answer. OpenAI says that with several relevant advertisers, "we'll select the one that is most relevant to your chat to show you first."
• Category rules change by market. The ad policy, version 1.6 of September 2026, states: "Ads for financial services outside the US are generally prohibited." It uses the same words for health services. Political advocacy ads are disallowed, and no ads run in mental and personal health conversations. Version 1.6 also lets OpenAI decline advertisers that conflict with its business interests or competitive position: in our reading, that matters if you sell anything OpenAI also sells.

Do Google AI Mode and Microsoft Copilot sell ads separately from search?

Not as a separate budget line. Ads in Google AI Overviews, AI Mode and Microsoft Copilot are bought through the Google Ads or Microsoft Advertising account you already run, even where the format itself is new.

Google AI Overviews. Google's help page on ads and AI Overviews lists 12 countries where ads can run inside the overview, in English, on mobile and desktop: Australia, Canada, India, Indonesia, Kenya, Malaysia, New Zealand, Nigeria, Pakistan, the Philippines, Singapore and the US. Text and Shopping ads from Search, Shopping and Performance Max campaigns are eligible; sensitive verticals are excluded, finance and healthcare among them, so an insurance advertiser's ad in Singapore can sit above or below an overview but not inside it.

Google AI Mode. At Google Marketing Live on 20 May 2026, Google said it was "testing two new types of ads, built with Gemini". With Conversational Discovery ads, "your ad answers a person's specific question"; Highlighted Answers place an ad inside an AI Mode recommendation list.

Neither has its own budget or report; Google's advice is to "build a strong foundation with AI Max for Search, AI Max for Shopping campaigns and Performance Max." Our Google AI ads guide covers each format.

Microsoft Copilot. Microsoft said on 5 March 2025 that ads in Copilot had "fully ramped in all English, French, and German speaking markets", with Spanish and Japanese to follow that month; we found no newer market list, so check your account. Showroom ads, a Copilot-specific format, were still a pilot in June 2025, with no general-availability post we could find.

Carolyn Chou, Director of Product Management at Microsoft Advertising, wrote on 27 August 2026 that ads within Copilot experiences are "driving 101% higher engagement relative to traditional Search ads." That is Microsoft's number, on its own definition of engagement and its June to August 2025 data. Our Copilot guide covers the campaign tools.

Reporting is the catch on both platforms. Google's page says ads in AI Overviews "are reported as Top Ads" with no segmented reporting; check your Microsoft reports before promising anyone a Copilot number. Only an experiment shows what AI placements add.

Will Gemini, Perplexity or Claude start showing ads?

Only Gemini looks likely, and Google has announced nothing. Perplexity executives said in February 2026 they do not plan to return to ads, and Anthropic committed on 4 February 2026 to keeping Claude ad-free.

Gemini: not yet. More than 1 billion people use the Gemini app every month, per Google's 11 August 2026 announcement. None of them sees an ad.

On Alphabet's April 2026 earnings call, chief business officer Philipp Schindler said, as widely reported, that a format that works well in AI Mode "would transfer successfully to [the] Gemini app." We read AI Mode as the test bed; budget reserved for Gemini ads next quarter is a guess.

Perplexity: stopped. It tested sponsored placements in 2024, then phased them out; in February 2026 its executives told the Financial Times they did not plan to revisit advertising. That is press reporting, not a Perplexity policy page, so re-check it each year.

Claude: no, by public commitment. Anthropic's post of 4 February 2026 says "Claude will remain ad-free".

For the three ad-free assistants the only lever is being cited; our AI Mode playbook covers that work.

How to budget a ChatGPT test and prove it worked

The map answers which AI chatbots have ads. Whether buying there paid is the harder question. Fund ChatGPT Ads as a capped test with an end date. Judge it, and any change to your Google or Microsoft matching, on incremental conversions: the extra conversions across all channels that would not have happened without the ads.

OpenAI now points the same way: on 5 October 2026 it announced geo-based incrementality experiments with Haus, Measured and WorkMagic, which it calls early stage.

Route 2 will take most of the first quarter's analyst time: nothing to switch on, much to measure.

1. Map eligibility by market. Extend the route table to every market you sell in, and date every cell.
2. Size the test before you cap it. Pull 12 weeks of daily conversions for the test market and its control; the widest six-week drift in the gap between them, with nothing running, is your noise band. If the test cannot plausibly beat it, run it bigger or longer, or skip it.
3. Cap it. A fixed share of non-brand search budget, six to eight weeks, a hard end date.
4. Hold out a comparable market. Singapore against Malaysia works if volumes and seasonality line up. Compare total conversions from all sources in your marketing analytics setup; our incrementality testing guide covers the design.
5. Write the rule down before launch. Name the lift that justifies scaling and who makes the call. Test Route 2 settings the same way.